The Southwest Biotechnology and Informatics Center would like to introduce
you to its bioinformatics computer system, darwin.
"darwin" is the name of SWBIC's
Beowulf cluster for high-throughput bioinformatics computing. A Beowulf
cluster is a collection of small computers (nodes) connected by a high
speed network. A host computer controls parallel processing applications.
Programs may be run in a truly parallel mode (i.e., one program divides
its processing among multiple nodes), or many programs may be run concurrently,
each on a different node.
SWBIC's cluster consists of 32 nodes, each with an 800 MHz Pentium III
processor and 256 megabytes of RAM. There are two 100 megabit network switches
that divide the load of communicating with the host computer. The host
computer is a dual-processor 800 MHz Pentium III machine with 150 gigabytes
of SCSI hard drive storage. The host and all the nodes are using
RedHat© Linux as their operating system.
SWBIC is using darwin for bioinformatics research at New Mexico State
University and for the bioinformatics services provided by the SWBIC web
site. As a benchmark, darwin can BLAST (blastp) a 400 amino acid protein
the NCBI non-redundant protein database (600,000+ sequences) in 24 seconds
(elapsed time) on a single node. This equates to the capability of completing
4,800 such BLASTs in one hour, using all 32 nodes.
|The dual-processor host computer
||100 megabit network switch (seen
on top of two nodes)